AI ยท #43
MedMon & Sentiment Analyzer AI
AI & Backend Developer (Solo)
๐ 91.38% Test Accuracy ยท F1 Macro ~0.896
Production (Internal)MedMon is an end-to-end AI-powered media intelligence platform that scrapes national Indonesian news and classifies sentiment using a fine-tuned IndoBERT model. Achieved 91.38% test accuracy and F1 Macro ~0.896 across 3 sentiment classes (POSITIVE / NEUTRAL / NEGATIVE). Fully automated the pipeline from scraping to dashboard display for PT ASABRI's Corporate Secretary division. The system includes real-time dashboard visualization, automated alerting for negative sentiment spikes, and historical trend analysis. The ML pipeline was built using Python, HuggingFace Transformers, and Streamlit for the dashboard.
Documentation



Have a similar project?Get in TouchCV / Dossier